Chinese phonology : an attempt to discover the sounds of the ancient language and to recover the lost rhymes of China

Waum
PLi2o!

V493

Gornell University Library
Ithaca, New York

CHARLES WILLIAM WASON

COLLECTION
CHINA AND THE CHINESE

THE GIFT OF
CHARLES WILLIAM WASON
CLASS OF 1876
1918

All books are subject to recall after two weeks
Olin/Kroch Library

DATE DUE

GAYLORD PRINTED IN U.S.A.

Cornell University Library
PL 1201.V93

“iii
3 1924 023 354 248 sw

24

Cornell University

Library

The original of this book is in
the Cornell University Library.

There are no known copyright restrictions in
the United States on the use of the text.

http://www.archive.org/details/cu31924023354248

CHINESE PHONOLOGY,

AN

ATTEMPT TO DISCOVER THE SOUNDS
OF THE
ANCTENT LANGUAGE
AND

TO RECOVER THE LOST RHYMES

OF CHINA

BY

Z. VOLPICELLI:

LATE BURSAR, ROYAL ASTATIC COLLEGE OP NAPLES,
HON. LIBRARIAN, CHINA BRANCH ROYAL ASIATIC SOCIETY.

On ne peut se flatter d’avoir le dernier mot d’une théorie, tant
quwon ne peut pas Pexpliquer en peu de paroles a un passant
dans la rue.

Chasles (Apereu- Historique suv 0 Origine et le
Développement des Uethodes en Geometric,
p. 115.)

PRINTED AT THE “CHINA GAZETTE’? OFFICE.

SHANGHAL 1896,

CHINESE PHONOLOGY,

ATTEMPT TO DISCOVER THE SOUNDS

OF THE
ANCIENT LANGUAGE

AND

TO RECOVER THE LOST RHYMES

OF CHINA

BY

Z. VOLPICELLI:

LATE BURSAR, ROYAL ASIATIC COLLEGE OF NAPLES,
HON. LIBRARIAN, CHINA BRANCH ROYAL ASIATIC SOCIETY.

On ne peut se flatter davoir le dernier mot d'une théorie, tant

qu’on ne peut pas l’expliquer en peu de paroles 4 un passant
dans la rue.

Chasles (Apergu-Historique sur 0 Origine et le

Développement des Méthodes en Géometrie
p. II5.)

一 一 一

TEL
UNIVER SS

fen FP RABY
PRINTED AT THE “CHINA GAZETTE”’ OFFICE,

cine 1896,

Wacom
PLIaol
V498

W. 1168

INTRODUCTION.

HE publication of this essay is mainly due to the

encouragement of Mr. A. von Rosthorn: during a
~ discussion on the second series of Rhyme-Tables contained
in the introduction of K‘ang-hsi’s Dictionary, I commu-
nicated certain views which I had long held on an
obscure point in those important tables. My opinions
were considered by Mr. von Rosthorn of sufficient import-
ance to induce me to undertake an exhaustive enquiry
on the different dialects of China with the material
industriously collected by Mr. Parker and inserted in
Mr. Giles’ Dictionary. The first results of my studies
are contained in the following pages.

T have carried on my investigation on a novel plan,
but if philology is one of the Natural Sciences, it cannot
be excepted fr.m the general treatment accorded to the
class: the collection of numerous specimens and their
comparison, with the application of mathematics (at
least in the elementary form of statistics) become indis-
_pensable for the discovery of general laws underlying
phenomena.

Suanenal, 1896.

Go the dlemory

OF

STANISLAS JULIEN.

Parr I.

RESULTS OBTAINED BY FORMER
SINOLOGUES.

HE first author to treat systematically of the sounds
of the Chinese language was Marshman, in his
“* Dissertation on the Characters and Sounds of the Chinese
Language’’ published at Serampore in 1809. His views were
very advanced for his time, and he was the first to point
out the analogy between the Sanscrit alphabet and the
initials of the Chinese phonetic system: he studied
very carefully the sound tables in K‘ang-hsi’s
Dictionary, and he reproduced the native views on
the subject, only venturing to suggest the probable
existence of the sonants g, d, 5, etc., in the ancient lan-
guage. He also showed the phonetic analogies between
Chinese, Siamese, Burmese and Thibetan. It is a pity
that Marshman stopped at the first series of tables in
K‘ang-hsi, if he had pushed on to the second set of
Rhyme-Tables, which are more ancient and fuller, we
should doubtless have owed some valuable discoveries
to his keen intuition and clear judgment. His arrange-
ment of the phonetic elements of the Chinese language
is lucid and methodical, and well worth studying by any
one undertaking similar enquiries.

The next writer to furnish material for the study of
Chinese phonology was Julien in his “Méthode pour
déchiffrer et transcrire les noms Sanscrits, ete’. His object
was to reduce to a system the various characters em-
ployed by the Chinese for rendering Sanscrit sounds, but
incidentally his researches threw light on the ancientsounds

6

of the language. But we owe much more to him: the
method he employs of converging a knowledge of San-
scrit, of Chinese, and the information contained in the
old Buddhist alphabets, on a problem which had hitherto
baffled solution, is of the highest order of scientific
research, and may be held up as a model for investigations
on such subjects. I have the fortune of coming after
him, and am able to adapt his method to the solution of
a similar problem.

Dr. Edkins may be considered the author who has
treated: the subject most exhaustively. He studied the
dialects, the old dictionaries, Julien’s transcriptions and
K‘ang-hsi’s Rhyme-Tables for the old sounds, and from
these different points of view he was able to collect evi-
dence for establishing, almost in every case, the value of the
old initials, and of the nasal and consonantal terminations:
he also made valuable contributions to the theory of tones.
These were very important results for one man to achieve,
and his success probably encouraged him to advance
those theories in general philology by which he is better
known. These bold speculations have diverted attention
from the less showy but substantial work which he has
done for the philology of the Chinese dialects; but time
will bring a juster appreciation, and on the latter will
rest the solid basis of his future fame.

We have shown how far Dr. Edkins advanced, but
he refrained from investigating the ancient vowels, as
their “variations are much more complex” and “vowels
are the most evanescent parts of words, easily become
modified, and an exact orthographic representation of their
nicer shades cannot be obtained.” ”

(x) Williams’ Syllabic Dictionary, Introduction Sec. V. Old Sounds of the Chinese
characters, :

7

The next names which occur are those of Dr. Chalmers
and Parker, who have both accumulated vast material for
the students of Chinese phonology,the former by his arrange-
ment of the rhymes of the Shih-Ching has furnished data
for the study of the language at a very early period, while
the latter, by giving in Giles’ Dictionary, the pronunciation
of each character in nine different dialects, besides Corean,
Japanese and Annamese, all romanised on a uniform
plan, affords the means of investigating Chinese by the
methods of comparative philology, which hitherto had been
impossible, owing to the want of a common basis on which
to establish comparison. I have derived, as I shall explain
later, all my material from Mr. Parker, and this essay
may be considered as the first outcome of his laborious
researches on. the dialects of China.

It will be useful to resume what has been achieved
and state the present condition of the question.

The Initials. Their old value may be considered as
established in almost every case, because though Parker
refuses to admit the general existence of sonants, they
are supported by the analogy of K‘ang-hsi’s Rhyme-
Tables with the Sanscrit alphabet, by the Sanscrit-
Chinese alphabets quoted by Julien, and by their exis-
tence at the present day in several dialects of China
and in the Japanese and Annamese pronunciation of
Chinese characters. And what renders the latter facts
~ more cogent is, that as I shall show, these sonants appear
in modern dialects and languages, for characters which
are placed in K‘ang-hsi’s tables under initials which
correspond to Sanscrit sonants.

The Nasal and Consonantal Terminations in the old
language are also well established by the modern dialects
in the South.

§

The Vowels and Diphthongs. These are very little
known, as Dr. Edkins, the greatest worker in the field of
ancient Chinese philology, admits they cannot be exactly
represented.

Native Literature on the subject. »This has been little
studied, even the valuable introductory matter in K’ang-
hsi’s Dictionary, so easily accessible to everyone, has
been much neglected. Marshman only took up the first
set of Rhyme-Tables, which are based on a comparatively
modern pronunciation; the second set which go back
much further, have only been cursorily examined by
Dr. Edkins, who probably could not spare time, and
had to depend solely on his Chinese Lettré, who com-
piled lists of characters and compared them with the
“Kwang-yun” and the second set of Rhyme-Tables.”

(1) Dr Kiihnert has studied these tables, but though his views are ingenious his
knowledge of the language at the time was insufficient for such a task. Hemisunderstands
phrases, and his residence in the country was too short for acquiring adequate phonetic
material. (See Journal. Academie der Wissenschaften Phil, Hist. CXXII. Band’
Wein, 1890.)

Part IT.

METHOD PURSUED IN THIS ENQUIRY.

T now behoves me to show the method I have
employed in verifying the results obtained by my
predecessors, and in attempting to advance a step further
in the difficult problem of the ancient Sounds of the Chi-
nese characters.

Estarted from the principle that it was necessary
for an adequate solution, to bring to bear on the problem,
evidence afforded by ancient sources and modern dialects,
that this evidence should converge on the same phonetic
material, and that the latter should be so extensive as
to ensure the discovery of a law not disturbed by casual
irregularities.

To satisfy these conditions, I took as basis of my
work the second set of Rhyme-Tables (24 in all) in K’ang-
hsi. These tables can be traced back to the Sung Dynasty
and probably date earlier, about 1,000 or 1,500 years
ago, a period when the Chinese paid much attention to the
sounds of their language, and when they secured the
collaboration of Sanscrit scholars, trace of whose work is
discerned in the arrangement of the tables. This is the
period which has been generally studied, as Dr. Edkins
assigns his old sounds inserted in Williams’ Syllabic
Dictionary to 1,209 years ago: it is also the furthest
period to which we can extend our researches for the
present. Besides it probably represents the state of the
language before it had undergone great changes due to the .
Tartar conquests.

10

These tables are arranged according to 15 termi-
nations (they are said to be 16, but two are amalgamated
together), some of which are also subdivided into open
(Bq a), and closed (4 ) sounds, the latter being
differentiated from the former by the insertion ofa w as
Kuan from Kan, Kuen from Ken, etc. Some of the termi-
nations are single, either open (§9 1); or closed (4 14), in
which case they have only one table.

Each table is divided into four divisions, each -of
which contains characters in the four tones. The initials
are placed at the head of each table so that all the cha-
racters in one column have the same initial. The rhymes
are placed in a column at the left of each table, so that
all the characters in the same line have the same tone and
final. The complete table contains 23 columns (there are
86 initials, but 13 of these are in double series, 2 for each
column) and 16 lines: of course there are not characters
filling always each column or line, and in some tables
one or more columns or ,lines are destitute of characters.
This means that given theoretical combinations of initial
final and tone did not exist practically in the language. .

A glance at these tables in K’ang-hsi will make
the above summary description perfectly clear, and the
reader will find that the terminations (3), the initials, and
the tones are perfectly easy to understand, what will
puzzle him at first, are the four divisions disposed
horizontally in four tiers, one under the other. They have
nothing to do either with tone, initial, or termination
and their meaning is not apparent. .

Many years ago, when studying these Tables with
‘the valuable assistance afforded by the directions given
in the introduction to Williams’ Syllabic Dictionary by

11

Dr. Edkins, I came to the conclusion that, as they could
not mean any change in tone, initial or termination, they
must indicate a change in the vowel and that they
expressed the various vocalisations with a common ter-
mination. This would not affect the rhyme, as different
ones are given for each division. By a summary exami-
nation of the characters under each division, with their
pronunciation, according to the Southern dialects, I con-
cluded that they were used to express the simple vowels:
0,a,e,%, (as in Sir T. Wade’s Romanisation). ‘I did not
pursue the subject, because I thought the conclusion was
sufficiently proved, but latterly, as my opinion was chal-
lenged by a friendly critic, who desired complete” evi-
dence for such an assumption, I proceeded with the
following enquiry. _ :

In these tables, I had a phonetic material of over
4,000 characters systematically arranged according to
initials, vowels, finals and tones. Each of these characters,
at the early period I have alluded to, had been placed in
such a position that it had a whole series of other characters
similar to it in one of the above mentioned particulars,
Here I had ready formed classes of similar specimens
earefully collected and arranged by the Buddhist
missionaries, worthy scientific predecessors of their
modern Christian brethren, who had first given China
a system of recording and preserving sounds. I thought
I could not better undertake the study of Chinese phonology
than by taking it up where it had been left by our Aryan
kinsmen, those devoted bonzes, perhaps the ouly foreigners,
if weexcepta few modern Christian priests and missionaries,
who. have ever acquired a deep knowledge of this strange
language so replete with difficulties.

12

I determined to follow up the dialectal variations of
each class, according as I wished to consider, the initial,
the vowel, or the termination. For this purpose, all the
characters in the tables were written out, according to
initials in single column, on large sheets of paper, then I
searched for each character in Giles’ Dictionary, and
wrote out in 12 columns, by the side of it, the different
sounds assigned by Parker to the 9 dialects and 3
languages(1), Some of the characters (rare ones) could
not be found in the Dictionary, and others had not the
sounds marked down for all the dialects; but on the
other hand, some characters had several forms for each
dialect, and as duplicates or vulgar forms have paramount
value for philological purposes, I recorded them all. I
thus collected a vast material of over 40,000 sounds on
which to base my enquiry. It was a most laborious
work), which I do not think I could undertake again,
and which I was sorely tempted several times to give
up: butthe work already accomplished, which would have
been wasted if left incomplete, deterred me from such a
pusillanimous course.

Such a vast mass of material threatened to overwhelm
me, and it required patience and thought to render it
manageable, I had to sift and resift before I could condense
results, and bring them into a clear and compact form so
as to be easily shown and not occupying too much space:

@) Only in the case of Japanese have I departed from the sounds given by Par-
ker: he gives the modern pronunciation, which is a corrupted form of the older one
preserved in the orthography which I have followed. Taken in this light, the evidence
afforded by Japanese is of the highest order, as it is the only ancient pronunciation of
which we have a clear written record.

(2) Iam so much indebted to the patient industry of Parker that I feel disinclined
to find fault even in detail with his valuable work, but his frequent omissions of dialec-
tal forms, which are placed under another character (which is not even referred to by
number) have enormously increased my difficulties. :

18

Though my original object had been to investigate
only the value of the ancient vowels, I determined to
utilise the large material I had laboriously accumulated,
to verify the conclusions reached by my predecessors j in
the matter of initials and terminations.

As I had arranged the four thousand odd characters

according to initials, (an arrangement to facilitate the use
ef Giles’ Dictionary, which naturally is also so arranged)
it was easy to follow up the variations of each initial in
the various dialects and languages. The comparative
table of initials is the result of this first investigation.
‘The enquiry into vowels and terminations, by which
latter word I denote, throughout this essay, only the
terminal vowels (in case of a diphthong), nasals, and
consonants of Chinese monosyllables, and which together
with the medial vowel or diphthong constitute what is
generally called final, which latter word I also use
in the-commonly received sense, was far more difficult.
... Firstly, because I had to examine them together,
secondly because the vowel variations were far more
complex,” and thirdly because the phonetic material at my
disposal was already arranged by initials. I had to sift
and resift the whole mass of sounds, and by successive
tabular arrangements, I at last reached the results which
will be found further on. The results published are the
6th tabular arrangements which I had to adopt to be able
to mould the material into a manageable shape.

Now that I have explained the method pursued, I
can proceed with the investigation and the results it seems
to authorise.

(1) An initial seldom presents more than 6 kr 7 variations, while a final generally
gives 30 or 40 in the different dialects examined.

14

#

Part ITT.

PHONETIC ELEMENTS OF MODERN
DIALECTS. “

S the object of thig enquiry is to bring the evi-
dence of modern dialects, and of ancient phonetic
arrangements to converge upon the same facts, it will be
necessary to rearrange the former according to the plan ~
of the latter, so that they may be investigated in a parallel
way. I therefore shall arrange all initials according to
K‘ang-hsi’s order, and vowels, diphthongs and termina-
tions according to the same system.

It will be useful to premise some considerations on
the elements we have to examine. Chinese monosyllables .
by native philologers are divided into two parts, initial
and final. But for a closer investigation it will be
sometimes necessary to effect a further division. Chinese
sounds, according to our alphabetical abstractions, are
composed of two or three elements: either of a consonant
and vowel simple; or of a consonant followed by two
vowels, z,¢e, a diphthong; or by a vowel and a nasal; or by”
a vowel and a consonant (in the Ju Sheng of the Southern’
dialects), or adopting the technical meaning I give to the
word termination, a Chinese monosyllable is composed
either of a consonant and a vowel, or of a consonant, vowel,
and termination. There are apparent exceptions to this
rule, as in some cases the sound consists only of a vowel
simple, or followed by a nasal or consonant; no initial
being manifest; but if we adopt the views of Arabic:
grammarians, we shall consider the vowel to be preceded.

TABLE OF IN {TIALS OF CHINESE DIALECTS.

CantOtaes rived. sacoeceure sieves sean
了 本 本 和 Us esleid
HOGCHOW? cccaxesiaccwsdveceeacees 6
Wenchow............+-
Ning pOscossccvexcess 和
Yangchow ....... 区
Weihien .............. sd ate
Tengchow ...... slehSleafaicintee sleet
IRIAN Ses sees s tor vewove seas
Nanking ...... 0
下 人

stews

ou, (
Wan tONs 2 ciiscstcsiaanceseds ”
DBE S acsids caietevacestaies

开 00chow 5 cscessazesvavssiel “43
AWWGHCHOW snccranvemaenccny
INIMEPOnscicesicessayesiage| 后)
AAD RCHOW [onc ass a9
WOMMOl: sicsaivaxayoveeens
Pengchow isssavssaxseas
Kivkiang 二 sse
NE
POkAN Sis rnacsrédeseeswell 49

TABLE OF FINALS OF CHINESE DIALECTS.

DENTALS. SIBILANTS,
GUTTURALS. LaBIALS, ASPIRATES. | Liquips.
Simple。 Palatal. Simple and Palatal.
K K‘ G (Gh) Ng | TT! D (Dh) N| Ch Ch'C C' Dj | P PB (Bh) MF V [t's Ts! Dz 8 Z Tx Ver sz Tez Sr Sh Sz Zz] H WY s| L J dr B
Canton...... St) ” ” 99 pylon on 和 yn 19. 99 ” ” ” 90.0 <9) ”» 099 ” 23
Hakka ....44| 5) 9» »y | ee) n>» nn ee) ” ) ny» non” ” 22
Foochow .| 5, yp | 9 pylon yw » ” » yn ” 16
Wenchow...| ,, 5, » 9 loon » 7 nnn |r» y 9» fr 9 oD » 9 ” » yn nf 9 31
Ningpo...... » wn ” yo 97 yy} 本 99 9799 9797 7 1 9 9 9 9997 93 29 v 34
Yangchow...| 5, 5, ) 1 nlm oo» nn )) 1 » on ” ym 4 ” nonmn wv} » » | 24
Weihien ... yn» ” ” yoy yn} n oo» nn )，)) ae et ” ” n ON yy Ww WT HW es 25
Tengchow... )) 99 yn” yin oo» np ” )) ” 99 ” ” 9 99 99--.99-4-99 20
Kiukiang...| 5, 4 » |» yi» 9 won) )) yn» ” ” » wn MO] ” » | 22
Nanking …| ， n>» 六 | ) )) >” » on ” ” yom” ” » | 20
Peking .| 4) yo” nlm oo» no») no» non » ” yyy gy an to 21
© 'These are true cerebrals,
ong on om og oun OUD On | ang an am aig aai aing aung adng| éng tn em eng en em efi eing ein eif
Canton......0+ ” yn» ” y ) >” yn yn» PM 9 » WW 9
Hakka. .| 9) 99 » ) 1) 7 no” )， 7)
Foochow ......-] 4) ” ” ”» 99 )) ” ” | ”
Wenchow... aoe » ”
Ningpo.... ones ” ” ” ” ” ” ” ” ”
Yangchow..... ” ” ”
Weihien aeseeeoe )) oo» ” ” y»
Tengchow ..... ” » 7) » >»
Kiukiang Sedawe ) ” 4 ” ”» oo»
Nanking sess] 95 » 4 yn » ” ))
Peking .| » 4 » >” | 9

”
”
”

”

”)

?》
”
»

”
”
»
”
”
”

ing | ing in im ing iung ion iang ian iam iaig ieng ien iem ien |

” » ”
»

” ”

”

»

”

”

TS.
AgSPIRATES. | Liquips.
Palatal.
Tsz Tsz Sr Sh Sz Zz] HW YHs| L J dr R
” ” » » » 9099 ” 23
” yh! 上 M2 29 99: 9? 22
pny ” 16
” ” ” W 3h 3 93: 2). 3 31
»y 2 ” ” 99 99 ?9 ?) 9 ?9 34
” ” ” ypnmn pv] mn ” 24
» » » yn 0 ” ” 25
” yn yn 7) | 7 20
” yn 7 99 99 ” ” 22
” 2 4}. 99 oh) ” 20
7 a” 39% 32 ” ste 21
cerebrals,

”
”

”
”
»
”

” ”
”
”

”

”
”

”

”

TABLE OF VOWELS OF CHINESE DIALECTS.

On, 0, é, a, aea, e, 6, i, i, u, ii,
Canton... miarareiareratetevarets; Per IOC OCT ICIDCCE ee Eee ” ” ” » Ny ” ” ”
ee ere erreeereescceccccues eer eeccce eee ccc cecceeeee ” ” ” y ” ”

OOCHOW: ses iesiadavens see nurs eeasers’ 0 让 让 ” ” ”
Wenchow...... anaieseid stere's Bae Susie Gareeiee esa Geiss eels 本 ” ry , ” ” ” ” ” ”
Ningpo.. Bielole’ ia dieters 人 Seer tie se be 6 ” ” ” ” ” ) » ” ” ”
Yangchow etna taenieete: SR 可 4% 4 ” ” » ”
Weihien Sa SG iSSS 0 bin ai 8inie O10 oe niai8 ole WHEE Te wise DES DESC s ” ” ” ” ” ”
Tengchow seedte daucsbages sauce sieaesees i 5 a a ” ” ” ”
Kiukiang......... sre vaqeendccawunsees 5 ” ” ” ” ”
NAMING cesstvady pieiniss cass eve ddvetegacasovendeads 前 ” ” ” ” ” ”
二 EEC 全 ” ” ” ” ” ” ”

TABLE OF DIPHTHONGS OF

OU, oui oi | ao, au, aui, aiu ae, ad, ai | Go, eu, ei, | iva, io, ioi, ia, iau, iae, iai, ie, ieu, iei, ie, lu | ui |
Canton eccaeicexccitaoiacs ” ry) ” ” ” » |
Hakka eeeeeeeseosoeeeeoseose ” ” » ” 1) eke | i) | ” ” ”
Foochow ee )) 99 ” ” » 9 9 99 yn 9?) ’” ny» )) » 99 ”
Wenchow....... Labeease no » m) yy » ) 0 no»
INNIS Oy 65 sa 是 让 二 ” yw fo ” ” ”
Vangchow sccissseseees| 94 » ” ” » ” » ”
Weihien cee eeereescaserees ” nly ” >» non ) 9)
Tengchow oo » ” » ” ” 9) ” ”

了 iukiang .ee ” ” ” ” ” ” ” ” ”
Nanking ev ” n ” ” ”
Ee RIDS: sotciuiedvunaaneunte ” n ” ” ” )) 7 ae) yn ) 7) 1”

CHINESE DIALECTS.

TABLE OF FINALS OF CHINESE DIALECTS.

”

én em eng en em efi eing ein eifi| ing | ing in im ing iung ion iang ian iam jaiig ieng ien iem ie icin lung iun
yp vy 0 ” ”
» 99 ” ” ?》 a: » ” ” ” ”
| ” ” ” » »
” ” ” ”
” 9” ” ” ” ”
” ” ” ”
yo” ” ” : ” ”
” ” y ” ”
” ” 》》 ” of ”
” ” >) ” ”
| n>” ” ” ” an)

ung un un uing

”
” ”

»

”

ting Un tian tien iie fein ting tin tik tung

’)

”

”
”
”

» 》

”

”

”

”

”

”

”

Ong on On 6ting Gin
”

” ”

» ”

loa, UO, tie, tiel, iil | of de

”

?9

”

n

”
”
”

15

by a Hamzeh (corresponding to the Spiritus Lenis of the
Greeks) or in other words, by a slight effort to emit the
breath which is a rudimentary consonant(, Thus such
apparent exceptions will be included in the general rule.

-I shall now commence my analysis of the modern
dialects of China, by giving first the elements which they
possess: initials, vowels, diphthongs and finals, arranged
according to the plan derived from Sanscrit, followed by
native philologers, and illustrated in the tables of K‘ang-
hsi. For the Southern dialects, I have dissected the
various syllabaries published by Mr. Parker, and for the
Northern (with the exception of ‘Yangchow”’ where I
have followed the same author) I have subjected to a
similar process the valuable sound table published by
Dr. Mateer in his ‘“‘ Mandarin Lessons.” I thus can give
a comparative view of 11 different dialects.

A glaiice at the foregoing tables will discdver at once
two important facts: the complete and harmonious series
of initials in the Wenchow and Ningpo dialects, and the
not less remarkable set of vowels, diphthongs, and finals in
the Canton dialect. In the latter, the vowels form a few
natural diphthongs, and join with the three” terminations
ng, n,m, to give a regular set of finals. 'The numerous finals
formed by the diphthongs commencing with 7 such as ia,
de, etc., are absent in Cantonese, and as this is the dia-
lect which agrees best with the ancient rhymes, we must
be led to suspect that these numerous diphthongs are
derived from simpler forms, in which an 7 has been in-
serted by euphonic necessities, consequent to the change of
value in the initials through phonetic decay.

(1) This conventional consonant will be marked *

@) T Jeave out the terminations 4, ¢,/, found in the Jz Sheng of the southern dialects,
because according to the views of the native philologers, whose system I follow,
they are but shortened abrupt forms of mg, 2, and m,

16

Parr IV..

PHONETIC ELEMENTS OF THE RHYME-
TABLES OF K‘ANG-HSI.

TNTITIATLS

The four thousand odd characters contained in the
above tables are arranged, as we have said, under 36
initials. These are divided into 9 elasses, which generally.
correspond to similar ones in the Sanscrit alphabet, which
thus becomes a precious auxiliary for determining their
value at the time when the tables were compiled. We.
subjoin the nine classes with remarks when necessary

st Class Gutturals K. K’. G. Ng. (same as Sanscrit.)-
2nd ,， Denials’? DNs (4. a op )
8rd ,, Cerebrals T(r). T(r).’ Dr). N@." (,, 5 » UE

This class was pronounced with the tip of the tongue.
against the palate as iftopronounce r. The dentals T.T’. D. N.
being uttered with the tongue in the above position, became .
respectively something between T, Ch, Tr-T’ Ch’Ts—D, Dj,
Dz, Dr—N, Ni, all of which sounds are found separately in
various dialects and in Annamese®),

@) I adopt the system of indicating cerebralisation by an yin brackets, beeause it
shows the position of the tongue: of course the ~ must not be trilled. Dr, Edkins is
wrong in considering the characters of this class’ as palatals; Julien’s alphabets show
they were u ed for Sanserit cerebrals.

(2) These sounds can be paralleled by the Syllabarium Ratatarum of Thibetan
(given by Giorgio) and by certain sounds in Dr, Mateer’s Sound Table in his Mandarin
Lessons,

L © @ N ld dod GW! SL SL 2a g Z
ce ee ce (74 a AL A M ce ce ce 《6
ML.(OL Od ON dd IWI@sL.(Os_ Meza (Og (DZ

LowWwadqN jdd GW] SL SL 7 8 Z
MHD SN UH H x A/T If
we HR MH eee Ww ew £ ww :
HRW & ARR ERE ee BW ke BwHamee
of ae HL ue HL 总
"SSYITO IST "SSVITD ae “SSVI ne “SSVIO, ons my 多 2

“S'IVIJINI SISH-ONVoN JO ANIVA ONIMOHS HIgVTL NOLE TANS

17
4th Class. Labials (strong) P. P’. B. M (same as Sauiscrit.)

Sth ,, 53 (weak) F. F.’? V. W. (absent.

This class is generally derived from the preceding 4th class

by the insertion of a 2:
6th Class, Sibilants Ts. Ts’. Dz. S. Z. (partially repre+

sented in Sanscrit}

wild

7th ., (cerebral) Sibilants Ts(r). Ts(r)’. Dz(r).S(r). Z(r)

(absent in Sanscrit.)

This, like the third class, was pronounced with the tip of.
the tongue against the palate, as if to pronounce 7, and the.
sibilants Ts. Ts.’ Dz. S. Z. by cerebralisation became some-

thing like Ter. Tsr’, Dar. Sr. Zr.

8th Class Aspirates. Hh. H.*. Y.

The sounds of this class, with the exception of Hhand Y, |

are very difficult to determine, and I give them with
diffidence as I am unfortunate enough to disagree with such an

os
a>

authority as Dr. Edkins.

9th Class L. Jr.
For the second, I prefer the sound Jr. toJ commonly given,
because the class would correspond to the half vowels ri and 7
of Sanscrit, and because Jr. may more easily change into
N. Jr. J. which occur so often in dialects and in Japanese.

(1) See note-on preceding page.

18

The 8rd, 5th, and 7th classes are not marked as
separate ones in the Tables under consideration, but are
placed respectively under the 2nd, 4th, and 6th classes.
But they can be no confusion, because K‘ang-hsi gives
directions for discriminating between them.

We have already stated that each table of K’ang-hsi
is divided into four Divisions, each containing four lines

of characters in the different tones. Now that dictionary

teaches that characters with initials of the 2nd and 6th
classes can only be found in the I and IV Divisions, while
characters of the 3rd and 7th classes are only found in
the II and III Divisions.

Characters with initials of the 5th Class are found
only in the III Division, and mostly in the Tables with
closed sounds (4 1).

A table is attached, showing how these different
initials have changed in the varions dialects and languages
which have formed the basis of the present enquiry. As
it would have taken too much space to give each dialect
separately, four groups have been formed:

Canton, Hakka, and Foochow marked S; Wenchow,
and Ningpo marked W WN; Yangchow, Mid-China,
Ssach‘uan and Peking, marked CV; and Corean, J: apanese,
and Annamese marked Z. The latter have often been
separated by a hyphen.

To show the relative frequency with which sounds
-occur in a group of dialects, 1 have arranged them in a
regular succession, those most commonly found being placed

1 Schlegel and Kiihnert both mistake the meaning of a simple phrase of K’ang-hsi
giving directions for this class, which would have become evident if they had criticaliy
examined the Rhyme Tables. Much learning was wasted in supporting the two mis-
taken views. (See Journal Akademie der Wissenshaften CX XII, CXXXI).

COMPARATIVE TABLE SHOWING DIALECTAL CHANGES IN K‘ANG-HSI’S INITIALS.
L.
KANG-HSI. 8. W.N. C.N. CorEAN—J APANESE—ANNAMESE.

Ksweiecinseiesav'e's ebadeuanine K. K, Ch, T’sz, Dj, Hs, Y, K, Ch, T'sz, Dj, Hs, Y, K.
K‘.. ese K, H, Ng, Y,= K, Ch, dj, h, g, hs, K, Ch, Ts, Dj, H, Ng, Y, W. K, H.
人 K. G, Dj, ¢, d. Ch, K. K—G—G.
Ng Ng,Y，W,”* H Ng, Y, W, = H. Ng, N, L, j, jw. « W, Y, H—G, K—Ng.
cassia ilivasuedelelai eae T, ch. T, D, tsz, dz, ch, dj. T, ch, ts. T, d, sh, ch, hs, tr.
TD) sussaeseaess oa Ch, t, ts. Ts, Dz, t, dy ch, dj, c, 8, ts. Ch, Ts, t, tsz. Ch, t, d, tr, s, hs, sh.
人 T, ch. Ts, d, ts, ch. fie T, d, ch, j, tr.
WMD) sis sedan suseeseuinaeses Ch, ts, t. T's, t, d, ch, dz, dj, c. T, ch, ts, ¢. T. D, ch, dj.
D) aagtne ss Seauisgaceasabates T, ch, s. D, t, dz. 'T’, ch, ts, ¢. oy D, ch. dj.
DE)! iaisicieccasiadiesjetalscadss Ch, t, sh, ts, s. Dz, dj, z, j, ts, ch, tsz. Ch, ‘I's, sh, hs, t, ch, sz. D, Ch, Tr, f, 8, hs,
IN: wcabacderiiecesestgaccuse N, L, ng, y. N, ng, ny, %, }. fa Fee Bree N, D, T. j, y
及 (全 N, L, ng, DYy，y) N, ng, y; Jj, 2. N, lL. DoT yj
了 下 ,二 :his P, B, f, v, h. iE a | P—H, f—T, f
和 F, h, p. F. V， h, p. |B. F.
下 P, hf m. P，B，V， P. P, H, Bf, t.
1 F, h, p. F, p, h, b. F. P—H—F, b.
和 Pf, by im: B, Py vy. P; P—H, B, f—B, t, f.
让 F, p, h. V, f, b, w. K p—F, H, B—F, b.
OA M. M. M. M—M, B—M, j.
We suceses 和 M, Nv: V,m, w. W,m. M—M, B, f—V, f.
人 T's, ch. T's, ch, dz, %. ‘T's, ch, tsz. Ch—S, Sh, z, j —T, tr.
'Ts(r) Ch, Ts. Ts, ch, j, z. ‘L's, Ch. Ch—8, Z—Tr, Ch, t.
人 Ts‘ ch, s. Ts*, ch’, dz. Ts Ch‘. Ch—S, sh, z—T, tr, 3, h.
WS) acsisescsue te ssueecee' Ch, Ts, +. T's, Ch, dz, z. T's, Ch. Ch, s—S8, Sh, t—Hs, 8, t.
有 Ts, Ch, s, 182. Dz, Z, ts, dj. Ts, Ch, tsz. Ch,—S§, Sh, Z, Dj—T, Tr, hs.
Dat): scdisccstedecsddesaes Ch, Ts, Sh. Dz, Z, j, ts, 8, sh. T's, Ch, sh, s, hs. Ch, S—S, Z, J—T, Tr, hs, e.
SD 全 8, ta, sz. 8, hs, sh, ts. S, Hs, ts. S, ch—8, Sh—Ts,
NU Sh, 8, Ch, Ts. 8, sh, z, dj, ts S, Sb, hes. 8, ch—8, Sh—S, T.
Lisvesesseceveresitessveceess Ts, 8, ch, tsz. Z, dz, dj, j, ye Hy, 8, tsz, ch, y, 4. S—sh, dj, z—T, d, }.
LAD) sepinsisacessasssavasssces Sh, s, ch. Z, Dz, j, Aj. Sh, ch, ta, ha, 8, ch—Sh, z, dj 一 二 让 hs, j.
Tih asia inisssecsewswesseces H, h, f, w H, K, hs, f, w, y,* H,” He, K y. H, k—K, w, g—H, K.
Re H, k. x W, Y, h, k, ng. H, Hs, y, k. H, k —k, G, w, y—H, k, g, }.
er ecashaswadawsshe Yi Y, W,: YW, = ng, h. Y.W, = h—Y W—Y,W, i, ng, lu, n.
V ckosavihaisevaccotevaceccnes Y, w, = h, ch, ¢ Y, h, w, hs. Y, w, i, hy hs. YW: —YW:s —V,J,g.
Viswcies adie doelenae ows esaeee’ L. L, n. i D. NRY—R—L n,
Jr. ¥, J, ny ny, 1, ¥. Z, J, n, ng, dz, y. J,n, l, ng, y, erh, w.

N.B.—The sign = means

absence of initial.

NY: —N, J, z, sh—-N, N, Ny.

19

to the left in capital letters, while those found rarely are
placed to the right in smill type.

From the table of initials of the different dialects of
China, it appeared that Werichow and Ningpo possess a
very complete series: the present comparative table shows
that their initials correspond closely to the theoretical
initials of Kang-hsi, It will also be seen that the sonants
are supported by Japanese and Annamese.

VOWELS DIPHTHONGS AND FINALS.

It will make the subject clearer, and render compa-
risons more systematic, if inverting the chronological
order, I premise the phonetic elements of the ancient
language which I have obtained as the final results of my
investigation.

.The hypothetical reconstruction of these ancient
phonetic elements is given on the next page.

20

HYPOTHETICAL PHONETIC ELEMENTS
THE ANCIENT LANGUAGE.

VowELs.

a

o, 6, a, e, 6, i (perhaps short i and i) u; i.

DirHTHonas.

00, Oi, Ou;
ao, ” ai, au;
e€0, el, i, eu;
io, li, iu.

FInats.

oang, on, om, éng, én;
aang, ang, an, am;
eang, eng, en, em, éng, én, ém;
iang, ing, in, im, (perhaps in with a short i);
ung, ung.

OF

(1) I adopt this spelling because it is followed by Parker, but the sound is really the

Ttalian o s¢vetio.

(2) This is represented as to by Wade and Parker, but I cannot follow such an
extraordinary combination for a sound so naturally rendered by the Italian diphthong
ao which gives the two vowels uttered. As the vowel system of Wade is based on

Italian, the spelling of that language has a right to be preferred,

21

As will be seen at aglance, this system of vowels
diphthongs and finals, is very simple, coresponds to Can-
tonese and Hakka, and, as will be shown later on, agrees
with the 24 Tables of K‘ang-hsi, explaining each of them
without the occurence of duplicates which deface the
reconstruction of the ancient finals attempted by other
authors.

It is now necessary to compare these theoretical
elements, with those found in the dialectal variations of
the 4,000 characters of K‘ang-hsi’s Rhyme-Tables, in
the way we have already done with the initials. But as
I have already pointed out, it is a far more difficult task :
the variations of vowels, diphthongs and terminations are
so complex, that they cannot be reduced to a system,
compared and judged by mere observation, however dili-
gent and reiterated. The mind cannot master hundreds
and thousands of forms, and judge of their relative im-
portance. It becomes indispensable to have recourse to
figures, those convenient symbols which enable us to rea-
son about facts whose multiplicity would baffle us if
approached without their assistance, I decided to form a
series of statistical tables. .

To economise labour I omitted all the Ju Sheng sounds,
as they could be readily constructed whenever the sounds
in the other tones were ascertained. Whenever a final
possessed two tables, for open and closed sounds, I only
considered the former, because the latter could be deduced
as a corollary from the table of open sounds once it was
determined: after these deductions I still had over 22,000
sounds to analyse. This residium was dissected, and by
six consecutive series of tables, reduced to condensed
statements of the relative frequency with which any

22

given final sound occurs in any, of the Tables of K‘ang-
hsi, pronounced according to nine different dialects and
the three foreign languages. Up to the last set of tables
I kept the sounds divided into the same four groups
which I gave in the Comparative Table of Initials, but
the tables were too cumbersome and presented such diffi-
culty for printing, that I had regretfully to condense
them into a single statement for all the dialects and lan-
guages given by Parker. This curtailment hides many
important facts, the conservative tendency of certain
linguistic groups and the facile phonetic decay of others,
become neutralised and lost in the general mass. But
even with these disturbing causes, the action of general
laws is still apparent.

In the tables which follow, the Roman numbers at
the top of each column indicate the Divisions of the tables
I. meaning the upper one, and IV. the lowest. Thearabic
numerals in the different columns indicate the number of
times any final occurs, and to know their just relative
importance, they must always be referred to the total at
the bottom of each column: they are the numerators of
fractions whose denominator is the total. In another
columm, the same number may have a different relative
value owing to the difference in the total.

It may be objected that statistics are inapplicable to
such phenomena, because in philology exceptions often
are of the highest value, and irregularities may indicate
lost forms, but I do not pretend to give an absolute value
to my fignres, nor should I decide in favour of a vowel or
a diphthong by a simple addition like a political division
in Parliament. I use figures, because they are the only
symbols that can render manageable such an unwieldy

23

mass, they can condense evidence and render it clear.
The reasoning on such evidence is not affected by their
use, and can proceed with the customary canons of the
science,

My general principle for drawing conclusions from
the material accumulated and arranged, has been to
subordinate everything to the plan on which the Rhyme-
Tables of K’ang-hsi were undoubtedly constructed. I have
asked from figures only a clue to show the nature of the
arrangement, I have not followed blindly the indications
of an arithmetical majority ; whenever the Tables required
that a difference should be found between two finals,
I have taken guidance even from numerical minorities.

In the comparative study of the fifteen terminations
of K’ang-hsi, at first, I found much difficulty in keeping
some of them distinct; my hypothesis that the 4 Divisions
(四 ) corresponded to the four vowels o,a,e,7, seemed
to multiply the number of finals, and give a largernumber
than the nature of the language consented. Buta diligent
study of terminations, arranged according to natural groups,
enabled me to distinguish finer degrees of vowel
differentiation.

I shall therefore present my tabular material, and
the conclusions I draw from it, divided, when ever possible,
in groups of two or three terminations; most of these groups
had already been formed by native philologers, and may
either be found mapped out in K’ang-hsi, immediately
preceding the 2nd set of Rhyme-Tables, or may be surmised
by the ancient arrangement of the 24 Tables which was
different from that given in K’ang-hsi. This latter point
will be discussed later. Only in one case have I ventured
to form a group of my own.

1 I thus discovered the narrow vowels and the # series.

24

N.B.—In the Tables illustrating the Terminations, for brevity
and clearness, I omit the vowel in all but the first form; for the
successive ones only the terminal consonants or vowels are given.

Ist Termination.
(1st AND 2ND TABLES OF K’ANG-HSI.)

This has both open and closed sounds, and is contained
in the 1st and 2nd Tables of K’ang-hsi. It may be
characterised as a simple vowel ending, because out of a
total of 2,263 dialectal forms collated, only 3in the Wen-
chowand Ningpo dialectsendinafeeble nasal peculiar to
thatregion, Wemust now examine how the vowel varies in
the 4 Divisions.

I. Division
The vowel o is well established: a glance at. the Table.
will show that out of 730 forms 433 contain that vowel.

II Drviston.
Hen also the vowel a, given by my hypothesis, is
well established, as it appears in 509 out of the 646
different. sounds.

III Drvistow.
The hypothetical vowel e, is also represented in a
majority of cases, as it occurs either simple, or in a
diphthong 252 times out of the total 433.

IV Diviston.

If we take into account all the diphthongs containing
the letter 2, it will be found in a majority of cases, but if
we consider the diphthongs a, iau, ze, ete., as they probably,
are, corrupt forms of a, au, e, then the vowel e, is the one
represented in the majority of forms. This would bring
us to the same result as the III Division which is not

Ist TERMINATION.

- (Open Sounps).

I II II | IV
332 77 22 1
DOW cere esl!) sass
69 3 Al) |S ot
os 2 = aad
i ie ee

1 10 4 3
202 | 460 58 68
12 sia ees ues

al ce 3 2
13 2 1 3
4 a
6 19 | 141 81
eo 3 2 2
1 28 13
1 aes a
4 1 8
3 12 29 57
1 és 2 1
ace = ae 2
10 47 45 49
ne ius Led oe
8 3 72 | IAL
ze 1 4
2 1 a 1
fis oe 12 9
a 3 4
36 i 1 Be
3 5 0 oe
a we 1
1 3
730 | 646] 433 | 454

25

logical, therefore to differentiate the two last Divisions,
and bearing in mind that the vowel 7, (or y), appears
rarely in the three first, and in many cases in the IV
Division we may safely assume it as the characteristic vowel
of this Division.

N.B.—For facility of calculation, and in the present
uncertainty about the primitive forms, I shall consider
_ every diphthong as belonging to each of its constituent
vowels, and reckon it with both.

2nd, 3rd, and 4th Terminations.

These terminations are grouped together in K’ang-hsi,
and their similarity justifies such a proceeding, the
characteristic trait is the nasal ng: out of 4,472() sounds;
3,276 end with n, ng, or ng and 488 end in wu, most of
which are Japanese renderings of this nasal: n only claims
- 885 and other miscellanens terminations only 323 sounds.

2nd Termination.
(3RD anp 4TH TABLES oF 玉 'ANG-HSI)，
This termination has only three divisions, the first is
wanting.

II Division.
The vowel should be a: out of 606 dialectal sounds
238 contain a, 188 contain e and 7, (which I consider as
the mute ¢ of the French) and 144 o and é.

III Drviston.

This should have the vowel e: out of 536 sounds
only 101 give the hypothetical vowel, while ¢ appears
in 353.

(t) This total does not agree with the pilditlons of the totals in the appended Table»
because some sounds have been omitted in the latter,

26

IV DrvrsroN，
This should be 7: out of 795 sounds, 617 give

that vowel.
The vowel dis given in both IIT and IV Divisions, but

there is a larger average in the latter, while ¢ appears
more frequently in the former.

3rd Termination.
(5TH AND 6TH TABLES oF 玉 'ANG-HST).
This may be considered to have only two Divisions,
because the II. only contains 3, and the IV only 7
characters, exclusive of Ju Sheng.

I Dryision.

Out of 465 sounds 236 give o and é, as the latter
predominates, I consider the vowel to have been a narrow
0, such as exists in French and Italian.

ITI Drviston.

Out of 481 sounds, e appears in 92, and @ in 261. By
analogy with the preceding Div. this was probably a narrow
e, a8 also exists in French and Italian.

4th Termination.

(77H TABLE oF K’ane-Hst).
This termination only contains c'osed sounds, and
forms therefore a different series with the vowels u, and i,
(u +e).

I. Drviston.
Out of 565 sounds , 467 contain the vowel wu.

(1) The /w-Sheng characters have little importance, as they are often repeated in
different Tables.

2ND TER-
MINATION,

3RD TER-
MINATION.

4rH TER-
MINATION.

Cec eee corse eoes

Corecess cevcee

ee

ee

ee cee ccc cccees

eovers ccc cccere

Co ry

eee ee: cere eccce

ee

eee coeeereee

eee ccceecescoee

ee。 of

oe ercerevces

@eecerer coe veces

eeseevseeeeeeoe

eeeeeoeeeeesees

eeseesseeseesseo。

cere ccces

eseeooeeoeos

eeeseeeeeeeeee

(OPEN SoUNDS.)

Il

il

(OPEN SoUNDS)

(CLOSED
SoUNDS.)

Ul

I ill

89| 34| “iol ja9] “49
38 | 28 3| 40] 32
oh lt eee RIN oe aie

93] 26) 99] 31 8

“43 1
49 | 93
13| 23

0°
心心 ”

6
re i 下 由 os:
39 13 | 17 5 3
it ; 9 可
26 | 2 Ek. Se
23 11
2 |

"88

a 16
24 | 32
2

“Jo

606

“7

536

(

95

wp 7

1 1

6 24
2 ee

a

. 1

i 3

c 1

34
34

Ha

329

48 1

565

27

ITI. Drviston.

Out of 726 sounds we have 512 giving u, but we
have also 67 with 总 (which does not appear in the last
Division) and 87 with iw, which is an easy corruption
from ii.

5th, 6th, and 7th Terminations.

These are grouped together by K‘ang-hsi, and
though the last one differs from the two former, it is use-
ful to keep them together as they form a parallel to the
group we have just examined. :

The two first may be considered to have had the ter-
mination 27, because out of 3,464 modern dialectal forms
2,122 have this termination.

6th Termination.

(10TH AND 11TH TABLES or K‘anc-Hst.)
For uniformity it is better to commence with this
one.

I. Drviston.

Here the vowel o, is represented only in about 14°,,’
while a, occurs in more than 55°,’, of the sounds.

II. Drviston.
Here the vowel a, appears in about 74°/, of the
total.

III. Drviston.
Here the vowel e, occurs in over 54°,’, of the total.

IV. Drvision.
The hypothetical vowel 7, appears as first or only
sound in over 55°’, of the total: no account has been

28

taken of those sounds, where i appears as the second
vowel, because then it might be the z of the termination.

5th Termination. —
(8TH AND 9TH TABLES OF K‘anc-Hst.)
As thereare only 21 characters in the I Diy.and11 in
the II. Div., exclusive of Ju Sheng, and assome of these are

printed in smaller type, and are not found in older works,
we may confine our attention to the remaining Divisions.

* III. Division.

Out of 726 sounds, the vowel e, appears only in
237 nearly a third, while 7, occurs in 467, I think
the vowel must have been originally an 6 fermé of the
French phonetic system. This Div. is a parallel to the
IIT Diy. of 3rd Termination.

IV. Drviston.
Here the vowel 7, occurs in about 75°,’, of the total.
It may have been a shorter ¢ than the one of IV Div.
6th Termination,

7th Termination.
(127% Taste or K‘ane-Hst.)
In the II. Div. there are only 12 characters, and in
the IV. Div. only 18, and as there are no rhymes we may
safely neglect these two Divisions.

I. Division.
This should have the vowel u as the 7th termination
has only closed sounds: the hypothesis is borne out by

over 71 °/, of the modern dialectal sounds, as out of 721
sounds 514 contain this vowel.

61H TERMINATION.

5TH TER-
MINATION.

7TH TER-
MINATION.

ee or erccreccees

eee ceeeeccccces

Core eereecer ees

eee corer cccvos

ee

eco cee ceecescee

Cece er ccc cccene

ee

ee

eeereeeeoeo，

seeseeveeweseeo。

Ceerercceccccee

Coc cor ere ceccoe

Ce eres cccesrece

Perc er cor eer roe

Coro eerseccceoce

Cec ccceccces ccs

Coeeer ce ccescce

Cer ecer ee vccces

were cercescrene

Cece ceeevcece

ee

ooooooeee-

eeeeoeeeseoeeo，

weveceeercceo eee

Cover rcsccccce

oe erecscseoeccs

eeseeoeseeeesee。

(OPEN SRoUNDS.) (OPEN SOUNDS)
I ]I | TI IV | III | 1V I III
1 10 1 1 1 142 51
54 1 1 1
ll os ate se
4) .. ; ov 122 15
Se 1 了
上 | 1
‘5 ‘i 2 2 DEL, «aid a
1 1 6 4 1 5 ars |) kOe
sins ie 3
13 5 oe
1 |
3L 97 1 1 3 see 17 6
285} 196| 92] 102] 1 14 1
see 3 1 |
1 2
3 :
1 ens : se ‘ aia |
47 41 2 24 5 3 |
3 1 2
2 28 |
2 |
1 |
92 5 33 80 ul 13 | 7
54 10 15 58 60 40 1 | 4
es 1 2
3 ] bys 3 12 ae wee es
2 1 3 3 2 L 1{ 21
1 2) 37
en ie 人 2
4 6| 28 5| 123| 5| AT, 了
mit ti “a 13 oa ee
ae us |
| | |
20 21 OL 329 | 445 | 243 39
i 3 ¥ 3
: 1 . 4 42
2 a ve
1 ee wes
es ie 1 1 7
‘ 13 1 sis
Bal ces eee 3 1 3
2 15 19 25 7 工作
1 ss 1 3 ，
2 4 -
1 Ses see 3% o 386 | 281
14 1 1 1 1
1 2 i
ie 1 DP lg 17 | 249
1 es ‘a 1
cay 1
675 | 471 | 184 | 650 | 726 | 341 | 721 | 879

8TH TERMINATION.

(Open Sounps.)

lltp TERMINA-
TION.

(OPEN SounDs )

er

Pere eerere es ecccces

Pee cer sere ccresees

ae

eeooeeseesee'

er ee ee ere er

Ce es

Co

Oe cee eeereccesccce

Cee eee ces coeees see

ee ee a

Cero eerercoccerces

Co

Seen Se

Ce ey

wee cero eereeeecce

Pe ecer cores cccces

ee cece een ecesescee

ee

sevewovoveooesseeo

II Ill IV I Ill IV
as ae io 1
4 ee 19 9 1
1 3 S| sass 1
1 二 6 1
1
14 12 9 44] 134} 39
4 15 9
1
37 iis 6
284 | 236 57 4 PF ase a
3 | 30 4 1 9| 21 14
7S ee a a 2 eee =
- a | 2
_ Pl eee Jt ees
5 3 2
i 2
1 2
24 TH) sey 6
29 1 2
1 6
2 a 国际 | 1
41 | 107 184 10 11 12
7 2 | 3 ‘ies 1
7 4 15 1 8 8
3 4 1 Old aes 3 a
- 1 iy = “s
5 28 35 3 2
1 2 1 TT gs
可 1
3 天
sie oP 4 10
2 1 13
1 2 ee 2
4 9 7 7} 139) 149
3 1 6 2 73 76
ss ee 1 1
Sas 3 2 del) Seve
4 me ox ss
8 2 4 3 8 1
a 21 90 | 190} ... be 21
2 7 OT | | -82 6 7 10
is sas see 4 | ,-
2 3 32 | 49 2 4
2 2| 30) 48 1 4
5 10 26 1 3
ss is se
oa 1
. is 2 :
4| 33 52 1 5
1 3 2 5 8
ia ， 1 1
下 | 总
2
ae 2 8
sn 1 1
1 1 1 1
514 | 506 55) | 182 | 488 | 366

£9
III. Drviston.

This should be &%: it appears in 288 sounds nearly
a third, and though w, appears a little oftener, still we
may consider % as established, because there must have.
been some difference between the I and III Divisions.

The 7th Termination has no nasal attached, so that
like in the 1st Termination the vowel is also the final.

8th and llth Terminations.

K‘ang-hsi groups the 8th with the 9th, but it is.
wrong, and contrary to the primitive arrangement of the
Tables, I follow that and the similarity in the ending.

The 8thand 11th Terminations contain 3,383 dialectal
sounds, of which 2,266 end in n, 6 in m, 695 in ng or &,
and 416 in miscellaneous vowels and consonants. We
may therefore assume n, to have been the common
terminal sound.

8th Termination.

13mH AND 14TH TABLES OF 玉 'ANG-HSI,)

I. Drviston.

The hypothetical vowel 0, occurs only in a little
over 7°/, of the total, while a, appears in over ae ie

II. Drvisron.
The vowel a, appears in over 72°/, of the total.

III. Drviston.
The vowel ¢, is established by over 78°/, of the total.

30)
IV. DrvIsIoN，
The vowel 7 occurs in 62°/, but e or 6 appear in over

87°/, of the total (of course diphthongs like ce or yé ap-
pear in both classes)

lith Termination.

(18TH AND 19TH TABLES or K’anc-us!.)

I. Diviston.
The vowel 0, appears in nearly half the dialectal
sounds, but principally as ¢, which is the o stretto of the
Italian language.

III. Drvisron.
The vowel e, appears only in about 13°’, while ¢,
occurs in 50°/, probably it was a narrow e, like the one
in the III Division of 3rd, and 5th Terminations.

IV. Drvision.
The vowel ¢ is established by over 75°,’, of the total.
It was probably a short 4, to distinguish it from the one
of the IV Division of 8th Termination.

9th and 10th Terminations.

Out of 2,357 sounds 642 end in m, 896 in n, 461 in
ng, and 358 in miscellaneous vowels and consonants ; though
m is in a minority, we must prefer it, because it nowhere
appears in such numbers, and because it is necessary to
distinguish these two Terminations from the two we have
just examined ),

(1) Of course the termination m, is strongly confirmed by tHe Southern
dialects, and its existence in the old language well established by Dr. Edkins. It
is therefore interesting to test the accuracy of my method, on a conclusion already

known

Coe rerccerecees

Pere eereesceces

CO

eeeseoseeeeeee

ee er oseo。

Cee rcerscrencee

eee eeccrone eter

coer reroee

eee eee reescoree

Or ry

eee oe eee ees

Poo coer ercoeres

Cee ee

ee

ee

Ce eo 玉 这 太古 下 于

ee

ad

Ce

ee

ee

ee ee

ee

er

0

0)

eu ..

oe er

ee eer

Ce ry

eee rseecoresere

ee ereeccreccere

seeeoeeeeeeeeee

Oe oe

eeoeseeeeeeee

9rH TERMINATION,

10TH

TER.
(OPEN Sounps,
I II Ill 1V III
号， oe 1
10 Qe Miata ve er
31 5 10 3 26
e sa ‘e 8
1 ee ‘ 1
3 6 3 81
17 : 4 63
i a eee
29 21 5 15 ae
106 65 22 11 1
143 | 102 51 3 10
32 20 5 af 33
oe 1 2
2
1 3
oe ve ]
eee . baci.) ooe 3
20 13 7 6 1
16 15 2 10 ]
1 1 8 22 Ae
3 1 2 3 8
4 28 61 to | 5
ae 2 Vr 13} 8
18 4 4 4 18
f 4 6 HA. ys
= is tee 1 | 2
21 OF] hes 1 | 1
1 ] 25 27
xe 2 i ;
1 en
13
1 10 2 TO ths
_ 1 4 5 64
1 2 二 117
1 oor 2 1 103
1 9 10 25 a
4 3 4 13
. 1 6 7 ere
soe 5 27 28 os
1 10 40 50 0
2 23 30 (On) ses
rf 8 23 34 2
an 2 11 22 rae
ie 1 4 Bi) ses
1 1 ae ee I
= 1) iene a
1 2 22 30
4 生 |
1 :
I
475 | 872 | 413 | 505 | 592

31
9th Termination.

(15ta AND 16TH Tastes IN K’ana-nst.)
The vowel o appears in about 13°, while a occurs in
over 73°,’, of the total.

II Drviston.
The vowel a appears in about 68°, of the total.

III Drvisrow.
The vowel e occurs in over 60°,’, of the total.

IV Drviston.
The vowel ¢ appears in 59°, but ¢ in over 74°’,.

10th Termination.

(17TH TABLE OF 及 ?ANG-HSI.)

In the I Division there are only 2 characters, in the
II only 12, and in the IV only 16, exclusive of Ju Sheng ;
we may therefore confine our attention to the III Division.

III Drvisron.

The vowel e, appears in only about 10°, while ¢,
occurs in over 51°’,. It may have been a narrow e like the
one in the III Divisions of 8rd, 5th, and 11th Termina-
tions.

12th and 13th Terminations.

These are only apparently separated, because all
the characters of the former are found in the latter, either
in the table for open, or in that for closed sounds. We
shall therefore only examine the latter Termination.

32:
13th Termination.

(218T AND 22ND TABLES oF K‘anG-HSI.)

Out of 1,624 sounds 1,452 end in ng (or au, tau, etc.,
which are the Japanese equivalents for this nasal) 2in », and
170 in miscellaneous vowels and consonants, we may
therefore assume ng to have been the final nasal, as we
have found in the 2d, 3rd, and 4th Terminations. There
must, however, be some difference in the vowels, and as the
simple ones are exhausted, I suggest the hypothesis that
a termination ang was added to the four characteristic
vowels of the 4 Divisions, producing the finals oang, ang
(aang), eang, tang. Let us see how existing facts bear on
this hypothesis. . .

I. Drviston.

The two vowels of the hypothetical form, o and a
appear respectively in over 37°/, and 60°,’, of the total,
but as they commonly appear in this Division with all
Terminations, no stress can be laid on the fact.

II. Drviston.
Here the vowel a, appears in 55°’, of the total, but
it does so everywhere, and we cannot tell whether in this
instance it is long.

III Drviston.

The vowel ¢ appears in over 21°’, and a in over 65°”,
of the total: the last figure is important, because the vowel
a rarely appears in the ITT Division: with all other Termi-
nations it occurs from 1 to 30°,, with an average under
14°’... We may consider the hypothesis verified by- this

Division

13th TERMINATION,

(OPEN Sounps.)

OD .ee

BU ,oo

工 I III IV
3 6)... 二
98 24 47 oF
39 10 12 “6
36 11 5 1
6 27 3 “

43 3 2
1 4 | wn oa
284 58 | 170 13
4 1 6 ‘a
12 3 1 eres
52 25 22 15
2 = 7
1 waa
one 1 eee eee
10 4 86 28
‘a 3 53 28
1 30 | 105 71
es me - AY 19
2 40 | 14
| 同
“ 1
1
1
i aes 1
1 9 可
ae 1 : 天
工 | oo.
is 2
594 | 226) 608 | 196

33
IV Divisron.

The two vowels, and a, occur respectively in over 70
and 67°/,, not only, but the finals iang, tang, tau, tae
appear in over 53°/, of the total. The hypothesis is fully
borne out by this Division

14th and 15th Terminations.

These are not grouped together in K‘ang-hsi, and in
the old arrangem: nt they were also kept apart, but I think L
am justified in treating them as forming a group. ‘They
have in common a vowel termination, which with the
characteristic vowels of the Divisions forms diphthongs
It is very difficult to distinguish between the terminal
vowels of the two, but I think I am justified in advancing

what follows.

I consider that in the 14th termination an 0, was
added to the four characteristic vowels, forming the
diphthongs 00, ao, eo, io. In the 15th termination I
think a u, was added to the same vowels forming the
diphthongs ow, au eu, iv. The modern dialectal sounds
show that these views are probably correct.

14th Termination.

(23RD TABLE OF 及 ANG-HSI.)

I Division.
The vowel o, appears in over 43°/, but the vowel uw,
appears in over 60°/,.

II Division.
The vowel a, appears in over 57°/, of the total, the
- termination 0, appears in over 30°, but w appears as
termination in over 64°,’, of the total.

34

III Drvisron.

The vowel e, occurs in 28°’, while 7, in 61°)’, of the
* total. Terminal 0, is represented by 22°/,, terminal u,
by 68° "55

IV Drviston.
The vowel 为 appears in 72°,’, of the total, Terminal

,

0, represents 21°,,, terminal u, 67°/,.

15th Termination.

(24rH Taste or K‘ane-nst)

I. Drvisron.

The vowel 0, appears in over 42°’, a, and e, in
about 19°’. Terminal w appears in 67°/,, terminal o, in

or
4 08

II. Drvistow.

The vowel a, appears only in 24°,’,, while 0, or é in
30°’,. Terminal uw, appears in 76°/,; terminal o, in 5°,’,.
III. Drvisron.

The vowel e, appears only in 8°/,, while 2, in 30°’,
and 0, in 22°’,.
Terminal « appears in 77°,’,, terminal o, in 8°/,.

IV. Divison.

The vowel 7, appears in over 56°’,. Terminal w
appears in over 90°. Terminal o only in 7°/,.

(1) The tendency of the diphthong ax is to become 9. It is worthy of. remark
that K‘ang-hsi has no rhymes to this Division.

14th TERMINATION.

15ta TERMINATION.

ee ereeeescesesoeee

we ereecretecscscee

Percocet sceseees

eee ee ee

eee eceesecessor cee

eee recs eeeccecccce

Ce

er oo

eee ee

Peeves ceesessenes

Peer eee ee

CO

eee ererorceresecee

Decree cre cceseseee

on

ee

ee rcorerceresseve

wee rvereecececares

ee

ee

oo

ee

eevesvesweesow

(Oren SoUNDS.) (Oren SoUNDS.)
1 II III | IV I II EEL) 4 EV
51 37 27 30 14 1 11 2
88 11 4 5 162 21 111 14
sine ie she 上 wae oat
131 101 27 17 1 a 5 4
1 二 了 sea ne
1 : 一 aius She eee
; 1 eas aes
1 1 2 3 3: 4 5 4
; 1 33 5 30 5
2 18 iva 3 re 1 1
283 | 278 105 48 95 17 66 39
wae 1 3 2 14 3 as 1
eee 1 1 3 1
3 ae an
2 3 3 2 ney 下 3
2 30 45 56 58 12 24 13
ies 1 ; we 4 |) sak es
2 Se oP 5 5 os :
one oe 7 1 1 oo
2 3 ] Al 7 4 1
45 3 3 6 4 5 1
see eae 1 3 wie
1 ¢ a 2 4 L7 6
2 四
1
¥ on 11 2 有 1
2 3 45 55 7 8} 218] 165
ie 3 2 a ie 2
5 14 17 22 1
2 20 59 73 1 4
a secs 1 12| 12
sigs IO) ice 3 “ …
8 42 115 | 202 3 13 13
4 12 75 80 3 a 6 7
1 2 3 38 ies 1 4
eek 1 a
|
= ee oi eae a me 9 | 1884. VL
vg as ‘ am 2 cre
a 0 - 1 APC oe
1 > 1 5 = 1m ee
oe 4 加 cen? - és tee
2 a 2 or see
工 | 。。， we ‘ie ie
646 | 588} 579 | 678; 575} 111 | 725} 370

35

If we compare the same Divisions of the 14th and
15th Terminations, we shall find that though terminal
u, predominates in both, there is a much larger number of
terminal 0, in the former: this fact seems to support the
hypothetical distinction.

We can resume the results of the preceding reasoning
in a tabular form, in which the sign « indicates where the
theory has a majority, - where it has a minority, and o in-
dicates when a Division has no character or so few that
it may be neglected.

L Division. II Division. III Division. IV Division.

Ast. Ter., oe * a
2nd » 0 R re F
3rd yy « 0 - 0，
4th 5 0 “ 0
55, ，0 0 = *
6 » wm * me % Sd
7 » 9» ® 0 * 0-
57 ，= .
95 » 7 ae a
10:5; 9 9 0 - 0
dhe ，”， * 0 a %*
13 ,, ” 7 * 出
14,, ，- * = 出
15 ,, 99 水 区 村 *
#==6 7 6 7 total 26
-一 1 8 3 ,, 17

Twenty-six Divisions are in favour, and 17 against
the hypothesis, all doubtful cases I have considered ne-

gative.

36

The adjoining Table shows the Rhyme-Tables in the
old order, and gives the finals according to the hypothesis
advanced. The Tables thus appear as a Syllabary of ©
the sounds of the ancient. language.

I now may resume the arguments in favour of the
theory.

1st. The hypothesis resists the severe ordeal of being:
tried by the whole mass of phonetic specimeris, and it
holds good in the majority of cases, notwithstanding all
the disturbing influences of so many dialects, several very
corrupt.

2nd. Even in the Tables. where one or more of the
hypothetical vowels do not occur in the majority of cases,
it will be noticeable that they always have a tendency to
appear in the place assigned to them by my theory. Ifo
does not represent the majority of any I Div. it.will be
seen always to appear there strongest, and gradually to
dwindle away in the other Divisions. The same happens
with each other vowel: 2 if not occurring in a majority of
the sounds of any IV Division, is seen to appear principally
there, and only sporadically in the others.

8rd. The numerical evidence ‘is: still stronger if we
confine our attention to the Southern dialects, especially
Cantonese.

4th. It is also geil by the authority of Chiang-yung
and other native philologers, whose obscure meaning seems
to be explained by my hypothesis, Chinese philologers
not having an alphabet, lack symbols to represent abstract

(1) Chiang Yung (ir 永 ) says:
—FRIOIGFKR
=§ aust mM

Ths Rhym3-Tables of K’ang-hsi arranged in the old ordsr and

with the finals resulting from the present investigation.

Number of Table
4 4, 3% (Closed Sounds) ung, ting.

12

11

14

13

2
15
10

9

” 江 (Open
(Closed
» IE (Open
(Closed
” 5B ( "
» & (Open
(Closed.
» 3% (Open
(Closed.
» UW (Open
(Closed
» 4 (Open
” wp ( ”
(Closed

» @ (Open
(Closed

» @ (Open
(Closed.

» $f (Open

” in ( n

» 深 ( ，

» BEC»

”

Finals,

) aang.

)uaang, /

) éi, i

) wei, ui.

)u, i

) ol, ai, ei, 1.

) voi, uri, uel, ul.

) én, én, in, 上
) unyukno， uin.

) on, an, en, in,

) uon, uan, uen, vin,

) oo (3) ac, é0, io.

) 0, ae i.

) uo, wa, ue.

) oang, aang, eang, iang.
) uoang, uaang, ueang.

) éng, éng.

) (too few to be notice.)
) ang, eng, ing.

) ou, au, eu, iu.

) ém.

) om, am, em, im.

37

sounds, and it is almost impossible for them to communicate
such abstract conceptions to others at least by writing. If
they. succeed: in having clear views on phonology, they
cannot impart them, except verbally to their: immediate
disciples.

5th. It is confirmed by certain sounds in the Wei-hien
dialect as given by Dr, Mateer in his “ Mandarin Lessons.”
There, as I have mentioned, one finds the sounds T's r, Tsr‘,
for characters, which are found in K’ang-hsi’s Tables under.
_ cerebral sibilant initials, showing that in Weihien they
keep a trace of these cerebrals. In Mateer’s Sound Table.
these sounds occur only with the vowels a, and ¢, exactly
as they should do according to my hypothesis.

6th. The hypothesis agrees perfectly with the old
. arrangement of the Tables, it explains the meaning of the
~~ ge and “& (Termination and Division), and of the different
groups, and produces a harmonious system which agrees

- almost in detail and always in spirit with the present

system of finals in the Canton dialect.

7th. The suggestion that the Rhyme-Tables of Kang-

hsi are a complete Syllabary of the ancient language is

supported by every reason of probability and historical
analogy. We know that Indians collaborated at the
‘compilation of these tables, and a syllabary is the natural
~ idea for all people who have been accustomed to use an
alphabet. It is also the method pursued by the Indian
Buddhists and their disciples in other countries. In J apan
we have all the syllables of the language arranged (in the
g9 ju on), by vowels, in the following succession a, 1, U,€, 0,
which is the order of the Sanscrit alphabet: the same
~ system and order is found in Thibetan. Far from being
astonished at finding a syllabary shadowed out even now 1n

38,

the Tables of K’ang-hsi, we ought to be surprised if the
contour were hidden by the mists of time.

If the results of this essay have, as I hope, advanced .
our knowledge of the ancient phonology, and enable us :to .
distinguish clearly those ancient” Rhymes which the
Chinese have laboured for ages to discover; and have-been
obliged gradually to throw away as their knowledge
became dimmer, it is a conclusive proof that without the
early use of an alphabet and the clear notions it gives, it is
impossible to discuss phonetic questions. I cannot flatter
myself that, in a question which has occupied the keenest.
intellects of China through many ages, I have not been
perhaps preceded by others, but through the abseuce of
alphabetical abstractions and the want of their convenient
symbols, the ideas of Chinese authors have been rather hazy
and they were utterly incapacitated from communicating
their results to the world.

I close this essay with a table of Rhymes according
to the Kuang-yiin. These are not all given in K’ang-hsi’s
Rhyme-Tables, and when given, are often differently —
combined: they probably represent a pronunciation which
was already antiquated when the tables were compiled,
and only preserved in poetry, where obsolete forms are
maintaine'l in all languages. The results of the general
investigation on K’ang-hsi’s Rhyme-Tables were, therefore,
not alone sufficient to discriminate all the rhymes of the
Kuang-yiin. I had to supplement the conclusions arrived
at with the formulas given by native philologers for using
the fan-ch‘ieh system. The sounds given in the next
Table represent the results (carefully weighed) obtained
by these different methods.

0 和 一 一

Jet ji

RE HA | ms EKA
ung WR & Blan i) £ HB
(ung) ? & he 沃 in | & GR EB
ing Pon ea | fl RR
aang iL 8% B |io a |
i | 3 BE R eo | Hah
él | B B& ao | FF IR
ei | 2 ib 00 | oe BF SR
néi | t BO 0 | he RB
i & i uo | Ae sh 4K
(i)? .| BO OB a | Kk & Is
ua | aH 大 eang .| By 38 BR Mt
. g fiom .| #& @ we
[CPO D pe ang Rw mh
泰 (nang) ?...) @F HR BF Bw
_ ee lene | a
34 ing | Ff WT ze B
ai | 4B EY 2 be 2k R&S
uoi IK Ba 队 eng ws | SOE OB

= AG eu | i 有 A
a |m eee: a
人
i Fe lem] ee
an Ey | on BER th ©
uén 女 吻 Rw (uom)?...| Be He PA de
in Rm MR | ek RR
uen 元 bt GB
~~ agzea? |” Ww AR R hh
be aan” |, & & BR
on ZB H 04| 8 BB H
uon seaxl m i
uan W me BR | uem A tv # Zz

@) In these three groups it seems that the characters have been mixed
up, as in the same line one finds the vowels < and a,

N.£L.—Doubtful forms have been put in brackets,

FOR SALE,

At the Office of The China’ Gaaette,
SHANGHAI

KELLY & WALSH, Lo.,

SEANGHAIT HONGIKONG AND YOKOHAMA.

OTTO HARRASSOWITZ,

IEIPZIG。

KEGAN PAUL,

TRENCH, TRUBNER & Co., Lp.,
LoxpoN。